marine accident report september 2016 - dmaib.dk ventus - fire on... · marine accident report...

MARINE ACCIDENT REPORTSeptember 2016

UMOE VENTUSFire on 23 December 2015

of 45

This marine accident report is issued on 19 September 2016.

Front page: Picture of SES. Source: Umoe Mandal AS.

The marine accident report is available from the website of the Danish Maritime Accident In-

vestigation Board (www.dmaib.com).

The Danish Maritime Accident Investigation Board

The Danish Maritime Accident Investigation Board is an independent unit under the Ministry

of Business and Growth. It carries out investigations as an impartial unit that is, organizational-

ly and legally, independent of other parties. The board investigates maritime accidents and oc-

cupational accidents on Danish and Greenland merchant and fishing ships, as well as accidents

on foreign merchant ships in Danish and Greenland waters.

The Danish Maritime Accident Investigation Board investigates about 140 accidents annually.

In case of very serious accidents, such as deaths and losses, or in case of other special circum-

stances, either a marine accident report or a summary report is published, depending on the

extent and complexity of the events.

The investigations

The investigations are carried out separately from the criminal investigation, without having

used legal evidence procedures and with no other basic aim than learning about accidents with

the purpose of gaining and promoting an understanding of safety. Consequently, any use of

this report for other purposes may lead to erroneous or misleading interpretations.

The Danish Maritime Accident Investigation Board Carl Jacobsens Vej 29 DK-2500 Valby Denmark Tel. +45 72 19 63 00 Email: [email protected] Website: www.dmaib.com Outside office hours, the DMAIB can be reached on +45 23 34 23 01.

http://www.dmaib.com/

of 45

CONTENT

1. ABSTRACT ............................................................................................................................................................. 4

2. FACTUAL INFORMATION .......................................................................................................................... 5

2.1 Photo of the ship ............................................................................................................................................ 5

2.2 Ship particulars ................................................................................................................................................ 5

2.3 Voyage particulars ........................................................................................................................................... 6

2.4 Weather data .................................................................................................................................................... 6

2.5 Marine casualty or incident information ..................................................................................................... 6

2.6 Shore authority involvement and emergency response ............................................................................ 6

2.7 The ship’s crew ............................................................................................................................................... 6

2.8 Scene of the accident ...................................................................................................................................... 7

3. NARRATIVE OF THE ACCIDENT ........................................................................................................... 8

3.1 Background ...................................................................................................................................................... 8

3.2 Sequence of events ......................................................................................................................................... 9

Summary of the voyage from the shipyard in Norway to Bagenkop, Denmark ....................... 10 3.2.1

Departure from Bagenkop – the fire and the evacuation of the craft ........................................ 12 3.2.2

3.3 The wreck of UMOE VENTUS ................................................................................................................ 15

4. INVESTIGATION DATA .............................................................................................................................. 17

4.1 Description of craft’s certification and manning, design, firefighting .................................................. 17

equipment and emergency procedures ....................................................................................................................... 17

Certification and manning .................................................................................................................. 17 4.1.1

General description of UMOE VENTUS ...................................................................................... 18 4.1.2

General aspects of composite materials ........................................................................................... 20 4.1.3

Structural and functional fire protection on UMOE VENTUS .................................................. 26 4.1.4

Emergency procedures – fire and evacuation ................................................................................. 27 4.1.5

4.2 The time, origin, cause and spread of the fire .......................................................................................... 29

The time of the fire ............................................................................................................................. 29 4.2.1

The origin of the fire ........................................................................................................................... 30 4.2.2

The probable cause of the fire ........................................................................................................... 31 4.2.3

The spread of the fire ......................................................................................................................... 35 4.2.4

5. ANALYSIS ........................................................................................................................................................... 38

5.1 The cause of the fire ..................................................................................................................................... 38

5.2 The operation and management of the craft ............................................................................................ 38

5.3 UMOE VENTUS’ robustness towards fire ............................................................................................. 39

Structural and functional fire protection.......................................................................................... 40 5.3.1

Carbon fibre reinforced plastic sandwich bulkheads ...................................................................................... 40

The crew’s capacity to fight the fire ................................................................................................. 42 5.3.2

6. CONCLUSIONS ............................................................................................................................................... 42

7. PREVENTIVE MEASURES TAKEN ....................................................................................................... 44

8. APPENDICES .................................................................................................................................................... 45

of 45

ABSTRACT 1. On 23 December 2015, a fire broke out on the prototype surface effect ship UMOE VEN-

TUS. The fire started in the port side lift fan compartment and within 15 minutes after the fire

was visually detected the craft was engulfed in flames and drifted uncontrollably until it

grounded in the shallow waters north of the harbour of Bagenkop, Denmark, and was lost.

The fire was caused by insufficient cooling of the lift fan engine exhaust system, which ignited

the exhaust muffler and/or the compartment where it was mounted. From the lift fan com-

partment the fire quickly spread outwards to the bulwark and accommodation and inwards to

the adjacent diesel oil tank. The insufficient cooling of the lift fan engine was likely caused by a

clogged sea chest strainer. There were several alarms on the cooling water system during arrival

and departure from Bagenkop, but the importance of the alarms was not acknowledged by the

crew due to events that had unfolded during the preceding days, which had created a tolerance

towards safety critical alarms.

After the discovery of the fire, the crew had no other option than to evacuate the craft without

any attempt to fight the fire manually and/or by means of the craft’s fixed firefighting systems.

The overall aim of the investigation was to establish why a failure of the cooling water system

led to an uncontrollable fire that resulted in a total loss of the craft. The focus of the investiga-

tion was UMOE VENTUS’ robustness towards fire.

It was found that the master worked in an environment of distributed authority – between the

charterer, the ship management organisation, and the owners. In the continuous communica-

tion with the shore-based technical and commercial management, the master was subjected to

other forms of authority that challenged his perception of his own authority on board the craft

which affected his judgement towards the seaworthiness of the craft.

The accident illustrated that it can be problematic to change a ship’s construction from a non-

combustible material to a combustible material by designing equivalent solutions based on tra-

ditional functional fire protection strategies. It was found that the concept of building the craft

as a combustible carbon composite structure with a relatively low ignition temperature, com-

pared to steel, reduced the craft’s robustness towards fire scenarios that were not considered

during the design and approval of the craft. Thereby, the fire shows the necessity of rethinking

the entire concept of the interaction between structural and functional fire protection, fire-

fighting and evacuation when changing the underlying premise of having the ship constructed

in a non-combustible material.

In order to increase robustness of the vessel, the shipyard has implemented measures on the

sister ship currently operating in wind farms and on existing and future new builds.

of 45

Figure 1: UMOE VENTUS Source: Valling Ship Management ApS

FACTUAL INFORMATION 2.

2.1 Photo of the ship

2.2 Ship particulars

Name of vessel: UMOE VENTUS Type of vessel: Cargo ship Nationality/flag: Denmark Port of registry: Svendborg IMO number: 9758507 Call sign: OWNK2 DOC company: Valling Ship Management ApS IMO company no. (DOC): 5588065 Year built: 2014 Shipyard/yard number: Umoe Mandal AS Classification society: DNV-GL Length overall: 26.6 m Breadth overall: 10.4 m Gross tonnage: 233 Deadweight: 14.9 m Draught max.: 2.77 m (on cushion) Engine rating: 2x1,440 kW Service speed: 40 knots Hull material: Carbon Fibre Sandwich Hull design: Catamaran hull with an enclosed air cushion

of 45

2.3 Voyage particulars

Port of departure: Bagenkop, Denmark Port of call: Svendborg, Denmark Type of voyage: Coastal (national) Cargo information: None Manning: 3 Pilot on board: No Number of passengers: 1

2.4 Weather data

Wind – direction and speed: SW 12-14 m/s

Wave height: 1-2 m

Visibility: Good

Light/dark: Light

Current: NE 0.75 m

2.5 Marine casualty or incident information

Type of marine casualty: Fire

IMO classification:

Date, time:

Very serious

23 December 2015 at approximately 1235 LT

Location: Coastal waters

Position: 54˚45.0’ N 010˚39.0’ E

Ship’s operation: Departure, underway

Place on board: Lift fan compartment

Human factor data: Yes

Consequences: Total loss of ship

2.6 Shore authority involvement and emergency response

Involved parties: JRCC Denmark

Resources used: Fishing ship GI-BRI SG 92

Speed of response: 5 minutes

Actions taken: Crew from UMOE VENTUS brought on board the fishing

ship.

2.7 The ship’s crew

Master: Held certificate of competency STCW II/3 – master home

trade.

52 years old and from Denmark. He had been employed with

the shipping company for 1-2 years and had served on UMOE

VENTUS for approximately 5 months.

of 45

Navigational officer: Held certificate of competency STCW II/3 – mate.

44 years old and from Denmark. He had been employed with

the shipping company and had served on UMOE VENTUS for

approximately 1 year.

Able seaman (AB): Held certificate of competency as an able seaman.

34 years old and from the Philippines. He had been employed

with the shipping company and had served on UMOE VEN-

TUS for approximately 20 days.

2.8 Scene of the accident

Figure 2: Approximate position of the accident, port of Bagenkop, Denmark Source: Danish Geodata Agency, chart 142, and © Made Smart Group BV 2016, C-Map data © Jeppessen AS 2016

of 45

NARRATIVE OF THE ACCIDENT 3.

3.1 Background

UMOE VENTUS was built in 2014 as a purpose designed service craft for transporting service

teams of up to 12 persons to offshore wind farms, where it could act as a stable platform from

where personnel and equipment could be transferred to offshore wind turbines. The craft was

designed to be effective in adverse weather and wave conditions.

The craft was a prototype surface effect ship, i.e. a catamaran with an enclosed air cushion and

a twin water jet propulsion system, which enabled the craft to reach a speed of up to 40 knots.

The hull, various tanks and the superstructure of the craft were built as a carbon fibre rein-

forced plastic sandwich construction.

It was delivered to the owner in February 2015 and was put into operation in March 2015. Ini-

tially there were minor prototype problems that were solved on site by technicians from the

shipyard. However, the crew considered the craft to be an overall stable platform for transfer

of personnel and with good handling characteristics that made it function well as a wind tur-

bine offshore service craft.

UMOE VENTUS was in a charter in Norddeich from where the offshore wind farm Borkum

Riffgrund 1 was serviced. A ship management company acted both as agent for the owners

and manager of the craft.

of 45

The narrative about the accident is in two parts: A summary of the voyage from the shipyard in

Norway to the location of the accident in Denmark, and a narrative about the fire and the evac-

uation of the craft. The narrative is presented from the perspective of the crew of UMOE

VENTUS as the events unfolded. Statements of time are given in local time in Denmark

(UTC+1), unless otherwise specified.

See figure 3 for an overview of the time and location for the main events prior to the accident.

Figure 3: Overview of time and location of the main events prior to the accident Source: DMAIB and © Made Smart Group BV 2016, C-Map data © Jeppessen AS 2016

3.2 Sequence of events

of 45

Summary of the voyage from the shipyard in Norway to Bagenkop, Denmark 3.2.1

UMOE VENTUS had been at a shipyard in Mandal, Norway, from September until December

2015 mainly for repairing a damaged transmission, but other minor modifications were also

made on the craft. During December the crewmembers arrived at the shipyard for familiariza-

tion with the craft’s modifications and training in the use of the craft’s control systems.

On the morning of 19 December 2015, after all final tests had been made by the shipyard and

the approval by the classification society had been concluded, the craft departed from the ship-

yard in Mandal bound for Norddeich, Germany, where a few days later it was to be put into

service carrying service personnel to offshore wind farms. The voyage to Germany was to be

made through Danish waters and via the Kiel Canal because the weather forecast predicted

unfavourable weather conditions in the North Sea. The plan was to reach the Kiel Canal the

same day giving the crew an opportunity to rest.

Upon departure the crew consisted of two Danish deck officers and one Filipino able seaman

(AB). Additionally, there were nine passengers from the shipyard, sales representatives, poten-

tial customers and crewmembers from a sister ship, who were on board to be familiarized with

the craft. They were to be disembarked upon arrival at the Kiel Canal. One warranty engineer

from the shipyard was on board to act as a liaison between the ship and the shipyard should

any technical problems arise. He was to be on board until the craft had reached Norddeich.

Approximately eight hours after departure from Norway and shortly before passing the Skaw,

Denmark, the craft experienced technical problems with one of the craft’s vent valves for regu-

lating the air pressure in the air cushion. The shipyard deemed it necessary to divert from the

voyage plan and go to the port of Frederikshavn, Denmark, for repairs and spare parts from

the shipyard. Shortly after arrival all the passengers, except the warranty engineer, disembarked.

After the repairs of the regulating valve and subsequent testing had been completed, the pas-

sengers, except the warranty engineer from the shipyard, disembarked. Two days later, on 21

December 2015 at approximately 1600, UMOE VENTUS departed from Frederikshavn with

three crewmembers and the warranty engineer.

Shortly after departure, the starboard side main propulsion engine malfunctioned and another

regulating valve for the air cushion system was not functioning according to the specifications.

It was once again decided to divert from the voyage plan and go to the port of Nyborg, Den-

mark, and wait for a service technician to arrive from Norway. He was to repair the main en-

gine and monitor the propulsion system while en route to the Kiel Canal.

Shortly after midnight on 22 December, UMOE VENTUS arrived in Nyborg. The AB and the

warranty engineer went ashore to a hotel and rested, while the master and mate stayed on the

craft.

of 45

The following day, the main engine service technician arrived and repaired the engine. Other

necessary adjustments of the ship's air cushion systems were made by the warranty engineer

from the shipyard.

The following day, 22 December at midday, UMOE VENTUS departed from Nyborg with the

warranty engineer and main engine technician on board.

While southbound along the east coast of Langeland, Denmark, the crew tried to pump ballast

into the starboard side ballast tank, but no content could be observed in the tank. After a

while, the ship got a 4-7 degree list to starboard. A bilge alarm was shortly after activated in a

compartment under the starboard side lift fan engine room. The warranty engineer went to the

lift fan engine room compartment to open a man-hole cover to inspect the compartment (dry

tank) and found approximately 7-8 m3 of seawater in the compartment, which was adjacent to

the ballast tank. The crew tried to pump out the water using the bilge pump, but the water lev-

el did not diminish and kept rising. The crew were uncertain about where the seawater came

from and considered the likelihood of the craft losing buoyancy and foundering. The crew

decided to divert from the voyage plan and proceed to the nearest port which was the small

harbour of Bagenkop, Denmark, to assess the situation in the engine compartments and pump

out the seawater.

During the arrival, as the craft was passing the breakwater, there were several engine system

alarms including on the port side seawater cooling water system, none of which the crew rec-

ognized to be critical because focus was on bringing the craft alongside and stopping the in-

gress of seawater.

In the afternoon of 22 December 2015 at approximately 1530, UMOE VENTUS arrived in

Bagenkop and the main engine technician, who had completed repairs on the main engine,

disembarked the craft. Immediately after arrival, the crew inspected the compartments below

deck. The master and the AB found that a man-hole cover on a ballast tank had not been

properly fitted and that the seawater came from the open ballast water tank. During the after-

noon/evening the water was pumped out using the craft’s bilge pumps assisted by a portable

pump which was brought on board by a representative from the ship management company,

which was located in the nearby port of Svendborg.

During the same evening the crew established that the engine alarm system showed multiple

alarms. They showed it to the ship management representative and they agreed that the ship

was not in a stable and reliable condition. By then it was evident to the master that it would

not be possible to reach Norddeich within the specified timeframe, and therefore the port call

was cancelled. It was, however, not convenient to stay in Bagenkop harbour because it was not

a commercial port, it was remote and did not have the necessary facilities for the crew. There-

fore, the crew agreed with the ship management company that they should proceed to the larg-

er commercial port of Svendborg, Denmark, to stay during Christmas and order a repair team

of 45

from the shipyard which could attend to the multiple engine system alarms and see to the bal-

last system that had caused partial flooding of the dry-tank.

On the morning of 23 December, the master had a conversation with the warranty engineer

and insisted that the main seawater cooling system filters be inspected before departure. The

filters were inspected and found to be clean. The sea chest strainers were not checked because

it was deemed unlikely that they were clogged because the mesh size was large. The fault that

prompted the alarms on the seawater cooling system was not identified.

Departure from Bagenkop – the fire and the evacuation of the craft 3.2.2

Upon departure from Bagenkop, the bridge was manned by the master who stood by the star-

board side conning station. The mate and the AB were on deck handling the gangway and

mooring lines. The master let the ship drift into the harbour basin and waited until the crew

had finished his work on deck. As the AB and mate arrived on the bridge, the master moved to

the port side conning station and increased power on the propulsion and departed from the

harbour at 1209.

Suddenly a large number of alarms were activated, including on the main engine cooling water

system. The master immediately reduced power on the propulsion engines, and the mate went

below deck where he inspected the various parts of the cooling water system and found that

the cooling water inlet pipe was cold indicating that it was in working order. The warranty en-

gineer from the shipyard heard the alarm and came to the bridge where he observed that there

was low pressure (below 1 bar) on the seawater cooling system. The master suspected that the

seawater filter was clogged and activated the back-flushing mechanism on the sea chest intake.

A few minutes later a high temperature alarm was activated on the port side lift fan engine, and

the mate stopped it and left the starboard lift fan engine running. The mate went to the engine

room compartments again and reported that he sensed a smell of overheating.

The mate and the technician had a short conversation with the master about returning to port,

and they agreed that they had to return to Bagenkop harbour to inspect the various systems for

any malfunctions.

Approximately ten minutes later as the master turned the ship to port and increased the pro-

pulsion power, a large cloud of black smoke from the port side engulfed the wheelhouse. The

master immediately shut down the propulsion engines, looked out of the windows on the port

side of the bridge and saw 1.5 metre high flames above the gunwale on the upper deck (figure

4). Shortly after the flames were observed, a fire detector alarm sounded from within the ac-

commodation.

of 45

Figure 4: Port side of UMOE VENTUS. Source: Private photo

Area with flames as witnessed by the master

The fire spread rapidly, and the smoke made it hazardous to stay on the craft. The master and

crew quickly realized that the fire was so intense that it would be meaningless to initiate fire-

fighting – focus was on evacuating the craft as quickly as possible. He announced to the crew

that they had to evacuate the craft. He immediately made a distress call on the VHF’s channel

16 at 1236. The coast radio station Lyngby Radio responded to the emergency call and relayed

the distress message to ships in the area. Within two minutes two local fishing vessels respond-

ed and departed from Bagenkop harbour. At this point, the fire on UMOE VENTUS was in-

tense and covered the accommodation on the port side.

While the master was busy with the communication, the crew prepared to evacuate the craft.

The mate and the technician went to the deck and launched the life raft, inflated it and lashed

it to the aft part of the deck. When the master heard on the radio that two fishing vessels were

proceeding to the position of UMOE VENTUS, he and the AB took the immersion suits and

life-jackets and threw them from the bridge to the aft deck. The master brought an emergency

VHF so he could communicate with the approaching fishing vessels.

Before the mate left the craft, he went to the bridge in an attempt to activate the foam extin-

guishing system, but the smoke was so intense that he considered it too hazardous to enter the

wheelhouse. The attempt was therefore abandoned.

At the time of the fire, there were 1.5-2.0 metre waves, which made it difficult to keep the in-

flated life raft steady alongside. The master’s initial plan was to wait for the approaching fishing

vessels to get alongside UMOE VENTUS. While waiting for the fishing vessels, the crew

of 45

donned the immersion suits, whereafter the master and mate went to the forecastle to assess

the situation. By this time most of the port side of the accommodation was in flames and

through the windows the master saw flames in the passenger lounge. The fire quickly became

so intense that they decided to evacuate the craft before the fishing vessels arrived.

The crew tried to push the life raft from UMOE VENTUS, but the life raft got stuck under

the aft part of UMOE VENTUS that pitched in the 1.5 metre waves. Water gushed into the

life raft, but after several attempts the crew managed to free the life raft from the craft and it

started to drift away. At this point in time the accommodation of UMOE VENTUS was en-

gulfed in flames and smoke, and the entire port side structure of the hull had been destroyed.

A few minutes later, the life raft came alongside a fishing vessel from Bagenkop and the crew,

assisted by the fishermen, embarked the fishing vessel. At 1257, the master called the company

from the wheelhouse of the fishing vessel and informed them about the situation. The crew

was brought to the hospital in Svendborg, Denmark, for observation for smoke poisoning.

Approximately 20 minutes passed from the time the fire was discovered until the crew were on

board the fishing vessel.

of 45

Figure 5: Starboard side of UMOE VENTUS shortly after the grounding Source: TV2/Danmark A/S

3.3 The wreck of UMOE VENTUS

When the craft had been abandoned, it drifted in the area north of the harbour of Bagenkop

for several hours before it grounded in an area of shallow water north of the harbour. The fire

developed a large amount of smoke that drifted inland and towards the town of Bagenkop.

The police authorities therefore urged the citizens to stay indoors until the fire had been extin-

guished.

The wreck was located approximately 60 metres from shore and it was therefore not possible

for the shore-based fire services to extinguish the fire. The fire was extinguished when the

structure of the craft had been damaged to the extent that all buoyancy was lost approximately

12 hours later, and it was almost submerged in the shallow water (figure 5).

On 17 February 2016, the wreck of UMOE VENTUS was salvaged in several pieces. About

half of the craft was salvaged as smaller pieces of debris. The wreck was brought to the port of

Horsens, Denmark, on 23 February 2016 where an investigation of the wreck was conducted.

It was apparent that the fire and the exposure to the sea had caused a complete structural

breakdown of the craft. The fire had destroyed the accommodation, the main deck and most

of the port side hull. The starboard side hull had also suffered a structural breakdown, but

some sections of the bulwark were still complete. During the salvage, the remains of the port

side hull were placed on top of the starboard side hull (figure 6).

of 45

Figure 6: The two hulls of UMOE VENTUS on a barge after having been salvaged Source: DMAIB

Starboard side hull

Port side hull

During the inspection of the wreck the port side platform management server was retrieved,

preserved in freshwater and brought to the shipyard in Norway for inspection. Two memory

cards (micro SD) were removed for inspection. One of these contained readable data from the

ship's platform management system. The extracted data were read at the yard before being sent

to the supplier of the platform management system. The retrieved data include trend logs,

event logs and alarm lists on the port side of the vessel for the period from 18 December 2015

until 23 December 2015

The crew had initially seen the flames coming from the fire on the port side from the area of

the lift fan compartment (figure 4). The investigation of the wreck therefore focused on this

section of the craft. The port side lift fan compartment was found completely destroyed by the

fire. The starboard side lift fan compartment was structurally partially intact.

of 45

In the description of the craft, the DMAIB will only address the systems and equipment which

were affected in the accident, or which could have or did influence the accident. Consequently,

neither the main propulsion and the main part of the auxiliary systems nor the navigation and

communication systems, etc. will be described in detail in this report.

This chapter consists of two parts (4.1 and 4.2): A section that gives a general technical descrip-

tion of UMOE VENTUS and a section that describes findings from the investigation into the

fire on 23 December 2015.

INVESTIGATION DATA 4.

4.1 Description of craft’s certification and manning, design, firefighting equipment and emergency procedures

Certification and manning 4.1.1UMOE VENTUS was registered in Denmark. The craft had been built and equipped at Man-

dal Shipyard, Norway, in 2014. The vessel had been approved as a cargo vessel carrying 12

passengers and three crew members according to Notice B from the Danish Maritime Authori-

ty. As an equivalence to Notice B from the Danish Maritime Authority, DNV-GL classifica-

tion standards as a small high-speed service craft for operation on offshore installations, in-

cluding wind farms, was accepted in combination with additional requirements from the Dan-

ish Maritime Authority.

The craft had been surveyed at the shipyard by a surveyor appointed by the classification socie-

ty who also surveyed the craft on behalf of the flag State. On 5 March 2015, the craft was is-

sued with a trade permit from the Danish Maritime Authority as a cargo ship with the opera-

tional limitation of: “Trade within GMDSS area A1 and A2 max. 150 Nm from nearest coast”.

On the day of the accident the craft’s minimum safe manning requirement was two crewmem-

bers1 (one master and one mate). The craft was not required to have an engineer on board be-

cause the ship management company had signed a service agreement with the shipyard that

had built the craft to provide regular maintenance and repairs, and because the craft had been

designed with propulsion redundancy so that it would be able to return to port if one of the

propulsion systems malfunctioned.

According to the craft's safety management system, it was the duty of the mate to carry out the

daily work tasks related to the engine systems. Daily repairs and maintenance were made by

local sub-contractors hired by the ship management company. During operation, the master

1 Ordinary ship assistants were not required when the daily voyages had a duration of less than 14 hours

provided that the ship's crew could perform their duties in accordance with the provisions on rest periods.

of 45

Figure 7: Extract from general arrangement Source: Valling Ship Management

Lift fan Vent valves

Draught reduction

was in charge of the navigation and the mate was in charge of monitoring the engine systems

on the engine control panels located on the bridge.

UMOE VENTUS had two crew cabins and was manned with three crewmembers, which

meant that one crewmember had to rest ashore. This put a time restriction of 14 hours per day

on the operation of the craft because one of the crewmembers had to disembark the craft to

get the mandatory rest hours. When the craft departed from the shipyard in Norway, it was

deemed possible to complete the voyage to the Kiel Canal in less than 14 hours, and the crew

would thereby comply with the rest hour regulations.

General description of UMOE VENTUS 4.1.2

UMOE VENTUS was designed as a service vessel for offshore wind farms, intended for carry-

ing technicians and small equipment items to wind turbines. The craft was a high-speed surface

effect ship, i.e. it had rigid side hulls like a catamaran structure with flexible rubber sealings aft

and forward. Centrifugal fans provided air pressure in the space between the hulls and thereby

provided lift to reduce the draught (figure 7). The pressure in the air cushion was regulated by

vent valves located aft of the lift fans.

The craft's hull and superstructure had been made as a carbon fibre reinforced plastic sandwich

construction. The design philosophy of UMOE VENTUS was a craft that could transfer per-

sonnel and/or equipment from the craft to offshore wind turbines in adverse wind and swell

conditions of up to Beaufort 5-6 (8-13.8 m/s) and 2.5 metre significant wave height, while at

the same time being able to reach high speed and low fuel consumption.

of 45

Figure 8: Extract from general arrangement Source: Valling Ship Management

Lift fan engines Diesel engines for water jets

The craft was equipped with two independent diesel driven water jet propulsion systems and

two engines for the lift fans on the port and starboard side (figure 8). The power supply system

comprised two diesel-driven generators, one located in each engine room in the port and star-

board side.

The entire hull and superstructure on UMOE VENTUS had been constructed of a composite

material. This structure enabled the craft to have characteristics that a conventional steel con-

struction could not provide, e.g. in terms of weight. Composites are made from a variety of

materials that give the composite structure different properties in terms of strength and flam-

mability. Therefore, it is relevant to examine the general characteristics of composite materials.

of 45

Figure 9: The sandwich configuration compared with an I-beam on the left

General aspects of composite materials 4.1.3

Introduction

In the last 30 years, fibre composite materials have seen a growing popularity in a wide spec-

trum of different industries. Areas of application have first of all been aircraft and spacecraft,

but with a decreasing fibre material price of the most commonly used fibre types, composite

materials have eventually been applied on a larger scale in ships, cars, trains, wind generator

blades, offshore installations, etc. Common to most of these weight critical applications is the

need for reducing the weight of the structure to increase the strength-to-weight and stiffness-

to-weight ratios and thus obtain better performance and/or an increased loading capacity, and

in many cases also reduce the maintenance costs. With regard to the strength- and stiffness-to-

weight ratios, composites and especially sandwich composites possess superior performance.

Other advantageous properties are thermal and acoustic insulation, fatigue, corrosion and easy

manufacturing of hydro- and aero-dynamically superior shapes.

Sandwich composites

In order to utilize the material properties of the individual composite materials to the best

structural advantage, a sandwich configuration consists of two stiff and thin layers (the faces)

separated by a soft and light material (the core) (figure 9). The three layers are in most cases

glued together, thus forming two glue layers in the sandwich. The material choice and location

of the different materials in the sandwich can be compared to an ordinary I-beam, which can

be regarded as optimized with regard to the cross-sectional geometry (figure 9).

The advantage of the sandwich compared to the I-beam is that the optimization can be ex-

panded to be used as panels (bulkheads), resulting in a highly optimized lightweight structure,

whereas the geometrical cross-sectional optimisation in the I-beam is limited to the beams.

Figure 10 below is a photo of a bulkhead from the wreck of UMOE VENTUS. The photo

shows the sandwich construction with the face and core material. On the upper part of the

photo the remains of the fire resistant insulation can be seen.

of 45

Figure 10: Section of bulkhead from the wreck of UMOE VENTUS found on the beach north of Bagenkop Source: DMAIB

Core material

Insulation

Face (fibre and resin)

Face (fibre and resin)

The photo shows that the core material has been exposed to heat and is partly melted, thereby

deteriorating the structural strength and integrity of the sandwich construction.

Almost any material which is available in the form of thin sheets may be used to form the faces

of a sandwich panel, and today a vast number of possibilities are available, making it possible

to tailor the sandwich construction to actual demands. The face materials can be divided into

two groups: Metallic and non-metallic materials. The metallic faces include aluminium, stainless

steel, titanium, etc., whereas the non-metallic faces are dominated by the fibre reinforced pol-

ymer based composite laminates (short FRP), with fibres systems like glass (Glass Fibre Rein-

forced Polymer – short GFRP), carbon (Carbon Fibre Reinforced Polymer – short CFRP) and

aramid (Kevlar) fibres, with various resin systems, such as polyester, epoxy, vinylester, as well

as a range of thermo-plastic resin systems (see illustration in figure 11).

Figure 11: Basic material constituents of a composite polymer material

of 45

Figure 12: Examples of random, crimped and non-crimped fibre mats.

In order for a certain fibre and resin type to be compatible, the surfaces of the fibres are chem-

ically treated to achieve bonding to the resin.

Composite face laminates can be constructed and manufactured with different types of fibre

architecture and categorized into basically three classes:

• Random orientated short and long fibre mats/fabrics

• Crimped/woven mats/fabrics

• Non-crimped multi-axial mats/fabrics

Randomly orientated short and long fibre mats or fabrics consist of fibres laid out in the plane

in a random pattern with a relatively large distance between the individual fibres, and therefore

resulting in relatively low fibre volume fraction and specific stiffness and strength. Crimped or

woven mats or fabrics consist of fibre tows woven into a specific weave pattern (figure 12).

Even though crimped mats consist of long unbroken fibres, their disadvantage is the bending

of the fibres associated with the weaving, which reduces the compression stiffness and

strength. Non-crimped multi-axial mats or fabrics are the most advanced and highest perform-

ing fibre architecture of the three classes and consist of a number of unidirectional laminas

with straight fibre bundles, which are stacked on top of each other in different directions to

form a multi-directional laminate with certain resulting stacking or lay-up sequences, 90 and -

45 degrees symmetrically repeated with the –45 layers forming the centre plane of the lami-

nate. The layup sequence chosen will change the mechanical behaviour of the laminate, making

it possible to tailor and optimize the mechanical properties of a specific laminate to fit certain

requirements in the structure.

Crimped and non-crimped mats can consist of either a single fibre type or of two or more fibre

types forming a hybrid laminate, and in most cases one or more of the fibre architecture clas-

ses are used in a commercial fibre mat product to enhance productivity.

of 45

Figure 13: Fibre mats from the wreck of UMOE VENTUS Source: DMAIB

The face sheets on UMOE VENTUS consisted of mostly multi-axial non-crimp mats in vari-ous layup sequences; similar to what is shown in figure 12 (right) and in figure 13 below.

The fibre types used on UMOE VENTUS were mainly of carbon and fibreglass. The polymer

resin/matrix types used in combination with the above-mentioned fibre mats for the manufac-

turing of the face sheets were of either the vinylester or polyester type depending on the loca-

tion of the bulkhead.

Other types of non-metallic face materials have generally also been used in sandwich compo-

sites, such as plywood, veneer and even cement. Common to all candidate materials is that the

primary demands on the face materials are:

• High stiffness giving high flexural rigidity • High tensile and compressive strength • Impact resistance • Surface finish • Environmental resistance (chemical, UV, heat, etc.) • Wear resistance

of 45

Figure 14: Basic core material types

Figure 15: Core of bulwark from the wreck of UMOE VENTUS Source: DMAIB

In figure 14, the four main types of core materials are presented, the corrugated, the honey-

comb, the balsa and the cellular foam cores.

The corrugated cores are normally used in heavy industries such as large ships, but have how-

ever also found their way into the packaging industry. The honeycomb cores are to a great ex-

tent used in the aeronautical industry as they possess the highest performance compared to the

weight. The honeycomb cores are made of for example aluminium, aramid or resin impregnat-

ed paper, which is the cheapest version and seldom used for structural purposes. Honeycomb

is also produced in a large number of different geometries, but today the hexagonally (honey-

comb) shaped type (figure 14, middle) is the most popular. Unfortunately, structural honey-

combs are also very expensive and less tolerant of impact loads, which limits their application

to relatively protected structures. The balsa and especially the structural cellular foams possess

a good compromise between performance and price compared to the honeycomb cores. They

are more tolerant of localised loads acting on the sandwich component. Cellular foams are

therefore the favoured core type in maritime structures.

The sandwich composites used in UMOE VENTUS consisted of the foam cored type with a

range of different foam core densities and fibre reinforced polymer face sheet configurations at

different locations in the vessel, as shown in figure 14 (cellular or balsa) and as seen in the pic-

ture below (figure 15) from the wreck of UMOE VENTUS.

of 45

There are several foam core types on the market, but the most popular is the structural polyvi-

nyl chloride (PVC) foam. PVC cores are available in a wide range of densities and material

properties and may be used in both a ductile (linear foam structure) and a more brittle version

(cross-linked foam structure). However, the linear ductile version is slowly being replaced by

the styrene acrylonitrile (SAN) foam core type, which is more tolerant of high temperatures

and in general a better performing material for structural use compared to the linear PVC

foam. Other core materials are the cheap polyurethane (PUR), which can be blown in between

the faces in a liquid form to subsequently densify, the polystyrene (PS), the polyisocyanurate

(PIR), the polyether imide (PEI) and the polymethacryl imide (PMI), which is more expensive

compared to the PVC core type and enjoys increased popularity in the aeronautical industry as

an alternative to the honeycomb core types. The most important demands on the core materi-

als are:

• Low density • Sufficient stiffness to prevent decrease in thickness under lateral loading (a limited decrease in

thickness leads to rapid decrease in flexural rigidity)

• Sufficient shear stiffness to ensure unwanted out-of-plane shear deformations • Sufficient stiffness to prevent local buckling of the faces (wrinkling) • Sufficient shear strength to prevent global core shear failure under lateral loading • Sufficient thermal insulation

Application of sandwich composites in the maritime industry

In the maritime industry, composite materials and sandwich structures have been utilized since

the middle of the last century. In the beginning mainly in smaller vessels like pleasure boats,

but also in more high-performance applications, like power boats. The step towards larger

composite vessels was taken by the military, just as in the aeronautical industry. Civilian appli-

cations of large composite and sandwich vessels have mostly been oriented towards high per-

forming, competition, oceangoing sailing boats, yachts, smaller ferries and special operations

vessels, such as offshore wind turbine inspection vessels, built by a number of Scandinavian

shipyards primarily in Denmark and Norway. An example of the latter special operations ves-

sels is UMOE VENTUS.

From the above it can be concluded that the diversity and continuous development of compo-

site materials makes it difficult to generalize about the structural properties of a ship built in these materials,

and how robust the ship will be towards adverse events such as fires and collisions. If the vari-

ous bulkheads are constructed in a composite material that is combustible, then robustness

towards fire must be obtained by the use of added structural and functional fire protection

initiatives.

of 45

Figure 16: Extract from fire and safety plan UMOE VENTUS Source: DMAIB and Valling Ship Management

Forward Aft

Structural and functional fire protection on UMOE VENTUS 4.1.4

Several compartments in the craft were fitted with structural fire protection, i.e. some of the bulk-

heads and decks were insulated to ensure that energy (heat) from the fire could be contained for a

certain period of time. The lift fan engine compartment was insulated to an A302 standard and the

propulsion engine compartment was insulated to an A60 standard, as it can be seen on the below

extract from the craft’s fire and safety plan (figure 16). The semi-open lift fan compartment was

not fitted with any structural fire protection.

UMOE VENTUS was equipped with three functionally different fixed fire protection systems:

A sprinkler system, a foam system and a NOVEC 1230 (a third generation substitute for Hal-

on). The sprinkler system covered the accommodation areas, including the passenger lounge,

pantry, stairway and hallway. All of the accommodation bulkheads had been built using a carbon

fibre reinforced plastic sandwich construction and therefore relied on the functional fire protec-

tion to provide protection equivalent to an A-0 division (uninsulated steel construction).

Functional fire protection (e.g. sprinklers) is based on certain conditions being fulfilled, e.g.

proper maintenance, due activation and that it is being used for the designed purpose, which

brings a complexity to the shipboard system that passive protection, such as insulated bulk-

heads or steel bulkheads, does not have (see the analysis section for further elaboration on the

subject of using functional fire protection systems). The foam and gas systems covered the

2 'A' class divisions are bulkheads and decks constructed of steel or other equivalent material, capable of preventing the

passage of smoke and flame to the end of the one-hour standard fire test. They are insulated with approved materials such that the average temperature of the unexposed side will not rise more than 139°C above the original tempera-ture, nor will the temperature at any one point, including any joint, rise more than 180°C above the original tempera-ture, within the time: A-60 (60 minutes), A-30 (30 minutes), A-15 (15 minutes) and A-0 (0 minutes).

of 45

propulsion engine rooms on the port and starboard side, and the lift fan engine room on the

port and starboard side. The sprinkler system could be operated from the exit door on the

bridge and from the port and starboard side exit doors on the main deck. The foam system

could be activated from the bridge and from control panels located in the starboard and port

generator rooms. The gas system could be released from a control station located in front of

the accommodation on the main deck.

Fires on the deck area were meant to be extinguished using the two fire hydrants on deck and

two fire hoses mounted in two cabinets on deck. One firefighter outfit including breathing

apparatus was located on the bridge. Additionally the craft was equipped with 18 portable ex-

tinguishers located in various places on the craft. On the day of the fire the crewmembers used

none of the firefighting equipment.

Emergency procedures – fire and evacuation 4.1.5There were no legal requirements for the craft to implement an ISM system. However, the ship

management company had decided to design and implement a system to systematically manage

the risks associated with operating the craft, as the company did on other ships.

During the fire and evacuation of UMOE VENTUS, the master and crew did not actively use

any of the formal on-board procedures. This gave rise to an investigation into why the crew

did not find the procedures to be a useful resource for handling the emergency situation.

The ship’s safety management system (SMS) contained procedures for emergency prepared-ness for a variety of situations. On the day of the fire, mainly two procedures from the SMS were relevant: Procedure for fire preparedness (SMS doc. 8.4.0) and procedure for evacuation (SMS doc. 8.3.3) (appendix 1).

In the procedure for fire preparedness, there were three references to generic documents for

use in a fire scenario (appendix 1):

A reporting form for the master to use in a fire (8-0400B) (appendix 1).

A public announcement template (8-0400C) for use when a fire occurred, and the passen-

gers were to be alerted and informed about the situation (appendix 1).

A checklist for extinguishing fires in various compartments on the ship (8-0400A)

(appendix 1).

It was not stated in the reporting form (8-0400B) what the purpose of it was: An instrument

for gaining assistance from the shore organisation, or an instrument for giving an account of

what had happened after the recovery of the emergency situation? The questions stated in the

reporting form suggest that it was to be used as the latter, i.e. to report about what had oc-

of 45

curred on the ship. Due to the necessity of a speedy evacuation and facing a total loss of the

ship, the recovery situation was not reached. Hence, the reporting was not relevant to the crew.

The public announcement template (8-0400C) was to be used by the master or another crew-

member on the bridge to inform the passengers about an ongoing fire. From the template it

can be seen that it was intended to be used in two circumstances: When the passengers were

not aware of a fire, or when the passengers were aware of a fire. The announcement template

seems to have been designed for a larger ship where the passengers would only have

knowledge about events in their immediate vicinity. The size of UMOE VENTUS meant that

the procedure would have little effect because the passengers would immediately be aware in

case of a fire. On 23 December, there was only one passenger on board (the shipyard warranty

engineer) and he was involved in the events to an extent that the use of the template was su-

perfluous.

The checklist for extinguishing a fire (8-0400A) contained instructions for how to act if a fire

occurred on either the deck, in the accommodation or the engine room. The checklist con-

tained instructions to the master and the AB (to follow orders from the master or the mate),

i.e. the checklist communicated to the master, the mate and the AB. However, some of the

instructions in the checklist were not addressed to a specific person, e.g. “activate the fire

alarm” or “start combating the fire”. The checklist instructed the crew about which firefighting

equipment was to be used in different circumstances. It was stated that a fire on deck should

be fought using fire hoses and a fire in the accommodation should be fought by means of the

sprinkler system. A fire in the engine room areas was to be fought with a combination of the

foam and gas system.

The fire on 23 December developed so quickly and with such intensity that the crew deemed it

impossible to fight the fire on deck with the fire hoses and decided to evacuate the craft with-

out using the firefighting checklist due to a number of factors:

The contents of the checklist were based on sequential thinking, with little consideration

of the dynamic nature of firefighting, e.g. were the authorities to be informed before

commencing the firefighting, thereby losing time for the rescue helicopter to arrive? Or

would it be expedient to have the crew muster at a specific location before immediately

starting to fight the fire?

The procedure stated that the passengers were to be informed about the fire, but another

procedure stated that the passengers were only to be informed in certain circumstances,

i.e. the checklist and announcement template could be viewed as being in conflict with

each other.

Lastly, the checklist instructed the crew to follow the instruction in row 6 in the checklist

for the fire in specific locations, but row 6 stated “choose adequate means for extinguish

the fire”, which would not be considered a specific instruction in a highly dynamic situa-

of 45

In this section the following topics will be addressed separately: The time of the fire, the origin

of the fire, the probable cause of the fire and the spread of the fire.

tion. Row 7 of the checklist contained references to specific equipment and stated “pre-

pare the fire hoses” followed by “start combating the fire”, which was a very specific in-

struction followed by a broad instruction to do something that involved many considera-

tions and individual choices.

The evacuation procedure of the craft did not have a reference to a checklist, but referred to a

procedure about shore-based crisis preparedness and an additional reference to SOLAS3 chap-

ter III (Life-saving appliances and arrangements). The evacuation procedure contained various

action points for the master to address in an evacuation situation and described that it was to

be done in cooperation with the shore-based crisis staff. The procedure contained instructions

about the state of mind of the master and/or crew, e.g. “stay calm” or “act masterfully”, but

the procedure also referred to specific tasks, e.g. “neutralize panicky behaviour” or “activate

the evacuation alarm”. Invoking a certain state of mind in an emergency situation is not likely

to be effective in a situation that calls for specific advice or instruction. During the evacuation

the crew were preoccupied with evaluating the right time to leave the ship, which was not ad-

dressed in any detail in the evacuation procedure, and they were preoccupied with evaluating

the possibility of evacuating the ship via the approaching fishing vessel, which was considered

safer than using the life raft as prescribed in the procedure. The problems gaining a safe dis-

tance between the burning craft and the life raft were not addressed in the procedure.

The fire developed and spread so quickly that it was difficult to stay on the bridge where the

SMS was located and left little or no time to read through the material and fill out the forms.

Furthermore, the emergency situation was not manageable with the strategies described in the

SMS due to a number of factors that were related overall to the dynamic nature of the situation

for which the SMS was not designed. Furthermore, the fire was so intense and developed so

rapidly that it was not considered possible to fight the fire with the fire hoses or use any of the

strategies prescribed in the various SMS documents. This indicates that the SMS was designed

for a type and size of craft other than UMOE VENTUS and/or for situations different from

the one UMOE VENTUS’ crew faced on the day of the accident. The usability of the proce-

dures will be further analysed in the analysis section.

4.2 The time, origin, cause and spread of the fire

The time of the fire 4.2.1The time of the discovery of the fire could be established on the basis of the witness state-ments, the master’s distress call, the on board alarm/event logs and the craft’s AIS data. The quality of the retrieved AIS data has been deemed to be valid because the AIS transmissions

3 The International Convention for the Safety of Life at Sea.

of 45

were found to be consistent over several days. The data have been considered credible because they are concordant with other collected data.

There were no fire detection alarms prior to the crewmember’s discovery of the fire. The initial

smoke and fire was observed by the different crewmembers on the bridge and the deck area

within a narrow time span. The master was positioned at the port side conning station and was

in the process of turning the ship and increasing the speed when he saw the smoke and flames.

From the AIS it can be seen that at 1234 the craft increased its speed and initiated a port turn

and changed heading from an easterly course to a southerly course after which the craft came

to an almost complete stop a minute later. The distress call was received by coastal radio sta-

tion Lyngby Radio at 1236. The discovery of the fire was therefore at approximately 1235.

The exact time when the fire broke out is, however, uncertain because the fire could have de-

veloped for some time prior to the crew noticing the fire. If the fire started in the lift fan com-

partment while the fan was in operation, the fan would direct the flames and smoke into the air

cushion while cooling the area, thereby limiting the out-board spread of the fire. The lift fan

engine was, according to the event log, turned off at 1223. Only when the master turned off

the lift fan motor, would the flames be directed outwards from the hull and become visible

from the port side bridge window. The rapid development and intensity of the fire suggests

that the fire had been developing for a period of time prior to its discovery. The mate inspect-

ed the main engine rooms at 1220 according to the event log showing that the water tight door

was opened for the last time. After this inspection the mate reported smell of overheating. This

was about 15 minutes before flames were observed. It is a possibility that the fire was develop-

ing at this time without being noticed.

The origin of the fire 4.2.2In the crew’s recollection of the events, the fire was initially seen coming from the port side lift

fan compartment. There was no warning about the fire from the craft’s automatic fire detec-

tion system, which was not activated (from the passenger lounge) until the fire had spread to

the bulwark. The extract from the fire and safety plan in figure 14 shows that the lift fan was

adjacent to the main engine room and the lift fan engine room. The main engine room and the

lift fan engine room were protected structurally by an A60 bulkhead and an A30 bulkhead,

respectively.

It is unlikely that the fire started within one of the adjacent engine compartments and burned

for a prolonged period, while all the fire detectors in the rooms malfunctioned, and broke out

through the bulkhead without the crew noticing. It is, however, more likely that the fire origi-

nated from the lift fan compartment itself, which was not equipped with a fire detection sys-

tem.

In figure 17 below is a photo of the starboard side lift fan compartment, which was similar to

the port side compartment. Inside the compartment there were four components: the lift fan

of 45

Figure 17: Picture of starboard side lifting fan compartment on UMOE VENTUS Source: Private photo

Exhaust gas from lifting fan engine

Drive shaft to fan

Lift fan housing

Exhaust muffler

housing, the exhaust pipe from the lift fan engine, the exhaust muffler and the drive shaft to

the fan.

There were indications that the fire originated from the exhaust gas system because there were

several alarms on the cooling water system while the craft was arriving in Bagenkop and upon

departure from Bagenkop.

Therefore, there is a likely correlation between the loss of cooling water pressure, the high

temperature on the lift fan engine, as experienced by the crew, and the lift fan engine exhaust

muffler which was cooled by the central cooling water system.

The probable cause of the fire 4.2.3

Figure 18 below shows the layout of the lift fan engine exhaust system. On the right hand side

of the layout is the engine, in the centre an A30 bulkhead and on the left is a picture of the

exhaust pipe and the exhaust muffler.

of 45

Lift fan engine A30 bulkhead Exhaust pipe

Figure 18: Picture of identical starboard side lifting fan compartment on UMOE VENTUS Source: Mandal Shipyard, private photo and DMAIB

Cooling water

Lift fan engine

Exhaust muffler

A30 bulkhead

During normal operation, the lift fan engine4 had an approximate exhaust temperature of 520

˚C. The exhaust pipe from the engine was injected with seawater (blue line) from the outlet of

the lift fan engine heat exchanger. The exhaust gas and cooling water mixed and flowed down

into the exhaust muffler (a lift-type silencer) in which the sound was reduced. From the ex-

haust muffler approximately 0.4 m3/s of exhaust gas was discharged and mixed with approxi-

mately 40 m3/s ambient air and directed into the pressurized cushion.

The lift fan engine was fitted with integrated class approved automatic shut-down functions

for low oil pressure, high cooling water temperature and over-speed. In addition there was a

cooling water flow-switch alarm – alarms which were periodically tested. In a situation where

the cooling of the engine and exhaust was ineffective, it would automatically stop while

prompting a common alarm. This functionality had, however, been designed only for the en-

gine and not the exhaust system, which was from a different supplier and fitted on the engine

system. Therefore, the exhaust muffler could theoretically have been exposed to a lack of cool-

ing water for a period of time without activation of the engine's cooling water temperature

shut-down function. Another possible explanation could be that the shut-down function on

the port side lift fan engine malfunctioned or had been incorrectly mounted after testing. It

was not possible to inspect the components after the fire because they had completely melted

away.

4 Make: Scania DI13 78 M.

of 45

In appendix 2 is an overview of trend data retrieved from UMOE VENTUS’s platform man-

agement server. The trend data cover the day of arrival on 22 December and the day of depar-

ture from Bagenkop on 23 December. On the horizontal axis time (UTC) is depicted and on

the vertical axis various system values are depicted. The system values shown are the main en-

gine RPM, the seawater cooling pressure and the lift fan bearing temperature.

It can be seen that during arrival on 22 December at 1435 (UTC), the seawater cooling water

pressure at times decreased to below 1 bar which coincides with various system alarms, includ-

ing the cooling water alarms the crew got while approaching Bagenkop – alarms that were not

reacted upon because the crew believed that the craft was about to sink (see section 3.2.1).

During departure from Bagenkop on 23 December, several common alarms on the port side

lifting fan engine were activated due to the continuous low seawater cooling pressure. The port

side lifting fan engine was stopped at 1123 (UTC) whereafter the master tried to back-flush the

seawater filters (testing high suction). Thereafter, at 1133 (UTC), it can be seen that the bearing

temperature on the port side lift fan engine was increasing indicating that a fire in the port side

lift fan compartment had started.

It has not been possible to determine with certainty what caused the drop in pressure on the

seawater cooling system because the fire caused extensive damage to the ship structure and

equipment that was spread out on the seabed. The condition of the valves and strainers could

not give reliable information about the state of the system at the time of the accident. An in-

vestigation of the main seawater pumps did not indicate any significant wear or malfunction

that would cause a significant drop in the seawater pressure on the main cooling water system.

The warranty engineer inspected the seawater cooling system filter before departure from Ba-

genkop, but the filters were inspected and found to be clean. The sea chest strainers were not

checked because it was deemed unlikely that they were clogged (the mesh size was large 8

mm). It is, however, likely that the sea chest was clogged by plastic or other material which had

been sucked up into the sea chest strainer during arrival in Bagenkop, because the drop of

pressure affected the entire port side cooling water system. The master’s attempt to flush the

sea intake would not have cleaned the strainer, because the flushing function was designed to

clear open the main intake and not the strainer.

The port side lift fan engine cooling water impeller pump would not be able to run for a pro-

longed period without a water flow before malfunctioning – causing insufficient cooling of the

lift fan engine and the exhaust. An investigation into the starboard side lift fan engine showed

that it did not experience a loss of cooling water as the exhaust muffler was found to be intact

(figure 19).

of 45

Figure 19: Picture of starboard side exhaust muffler found on the salvaged wreck Source: DMAIB

The loss of pressure on the main cooling water system would not necessarily result in an even-

ly distributed loss of flow of cooling water in all the port side engines because the port side

main engine was equipped with a larger centrifugal cooling water pump, which would create

larger suction than the smaller impeller pump on the port side lift fan engine.

The lift fan engine exhaust muffler had been made by a sub-supplier and been approved by

Lloyd’s Register to a maximum operating temperature of 85˚C. If the flow of cooling water

was disrupted, the exhaust muffler would be directly exposed to the exhaust gas from the en-

gine at an approximate temperature of 520 ˚C. This could ignite the muffler and/or the deck

and/or bulkhead where it was mounted. The hose connecting the exhaust piping from the en-

gine with the exhaust muffler had been designed for use at temperatures of up to 180 ˚C and

could therefore also be ignited by the exhaust gases.

It was unclear at which temperatures the bulkhead would ignite because the bulkhead’s fire

resistance had only been tested to be in compliance with the relevant IMO resolutions,5 which

means that the tests were based on bulkheads with insulation. It was therefore unclear whether

the composite bulkheads could be ignited directly from the exhaust gas or if another fire was

necessary to facilitate the necessary temperatures for the bulkhead to ignite, e.g. a fire in the

exhaust muffler.

In order to establish the ignition temperature of the bulkhead in the lift fan compartments, the

DMAIB requested the Danish Institute of Fire and Security Technology (DBI) to conduct a

test of the ignition properties of bulkhead samples from the lift fan compartment and super-

5 IMO Resolution MSC.307(88) – (FTP Code 2010) and IMO Resolution MSC.61(67) – (FTP Code).

of 45

structure of UMOE VENTUS (appendix 3). Both painted bulkheads and bulkheads without

paint were tested.

The following is an extract from the test report:

“The conclusion on ignition temperature is that the painted panel can ignite in the smoke gas tem-

perature range of approx. 275-315 °C and the untreated panel in the range of approx. 330-370

°C”.

“Ignition temperature” is a non-fundamental parameter that should be used with care. Ignition oc-

curs when the right smoke gas temperature and the right smoke gas/oxygen ratio are present –

this is never the same in each test. Is either “correct” temperature or “correct” smoke gas/oxygen

ratio missing - no ignition occurs. Furthermore temperature is device dependent measurement”.

“These test results relate only to the behaviour of the product under the particular conditions of the

test, and they are not intended to be the sole criterion for assessing the potential fire hazard of the

product in use”.

The test indicates that the exhaust gas temperature (520°C) from the lift fan engine had the

potential to ignite the bulkheads in the lift fan compartment. This means that the heat radiation

from an overheating exhaust muffler and/or the connected hoses could have started the fire in

the lifting fan compartment.

The spread of the fire 4.2.4Due to the extensive damage to the salvaged wreck, it was not possible to establish with cer-

tainty how the fire had spread in the early stages of the fire. Therefore, it was difficult to estab-

lish the duration of the fire prior to the crew’s discovering it. The development of the fire in its

later stages can mainly be established from aerial photos and witness accounts.

Within a few minutes after the discovery of the fire, it spread to the bulwark and the port side

of the accommodation. Within 10 minutes there were visible flames inside the passenger

lounge. After approximately 15 minutes and shortly after the crew had evacuated the craft, the

entire accommodation was engulfed in flames.

of 45

Figure 20: Picture of UMOE VENTUS seen from above Source: TV2/Danmark A/S

Figure 21: Picture of UMOE VENTUS seen from aft Source: TV2/Danmark A/S

Figures 20 and 21 are aerial photos after UMOE VENTUS had grounded – approximately 3

hours after the fire was discovered by the crew.

From both photos it can be seen that the starboard side outer hull was almost intact. The en-

tire accommodation was destroyed and the centre deck structure between the two hulls had

collapsed inwards.

of 45

Figure 22: Extract from general arrangement plan UMOE VENTUS Source: DMAIB and Valling Ship Management

Diesel oil tanks

Origin of the fire

Figure 22 is a general arrangement drawing of UMOE VENTUS seen from above. Two of the

craft’s three diesel oil tanks were located below the passenger lounge and can be seen just next

to the port side lift fan compartment where the fire originated. The diesel oil tanks had been

completely destroyed by the fire. The part of the diesel oil tank which was not facing the en-

gine room was protected neither by structural (e.g. insulation) nor by functional (e.g. sprin-

klers) fire protection. It had been made of the same material as the lift fan compartment and

accommodation. There was approximately 5,000 litres of diesel oil on board.

Presumably, the fire spread not only outwards and upwards, but also inwards to the diesel oil

tank, which fuelled the fire in the deck area between the hulls and upwards through the pas-

senger salon in the accommodation. The fire in the central deck area weakened the structure to

the extent that the craft’s two hulls collapsed inwards (figure 21). As the deck area was sub-

merged in the sea, the fire was extinguished except in the two hulls (figure 20).

of 45

ANALYSIS 5.

The overall aim of the investigation was to establish why a mechanical malfunction of the cool-

ing water system led to an uncontrollable fire that engulfed most of UMOE VENTUS within

approximately 15 minutes after the fire was visually detected and resulted in a total loss of the

craft. The focus of the investigation was UMOE VENTUS’ robustness towards fire.

5.1 The cause of the fire

The investigation has found that the cause of the fire on UMOE VENTUS was insufficient

cooling of the lift fan engine exhaust system, which ignited the exhaust muffler (including con-

nected hoses) and/or the bulkheads where the exhaust muffler was mounted on the port side

lift fan compartment. Due to the extensive damage to the craft, it has not been possible to es-

tablish with certainty why the seawater cooling system lost pressure resulting in insufficient

cooling of the lift fan engine exhaust. A likely scenario was that the sea chest strainer was

clogged resulting in an insufficient flow of water to the central cooling water system on the

port side. The cooling water pumps on the port side main engine managed to create sufficient

suction to supply the port side main engine with cooling water, but at the same time also de-

prived the port side lift fan engine of a sufficient flow of seawater. The lack of cooling water

flow to the port side lift fan engine cooling water impeller pump caused it to malfunction and

completely stopped the flow of cooling water to the lift fan engine. The automatic shut-down

function on the port side lift fan engine malfunctioned, causing it to run while overheating and

without providing cooling of the exhaust system.

During arrival and departure at the last port, Bagenkop, there were several system alarms on

the cooling water system indicating that the system did not function properly. The crew’s and

warranty engineer’s fault-finding efforts did not provide any clarity as regards the source of the

alarms. The crew and the manager of the craft chose to continue the voyage from Bagenkop.

The decision to continue the voyage and not promptly respond to the alarms on the cooling

water system should be seen in the context of the events from the preceding days, which will

be described in the following section.

5.2 The operation and management of the craft

During the voyage from the shipyard in Mandal, Norway, to Bagenkop, the crewmembers had

experienced numerous technical problems and alarms that gave the crewmembers the impres-

sion that the craft was in an unstable condition. However, the master did not consider the craft

unseaworthy – mainly for two reasons:

Firstly, the master relied on the warranty engineer to assess and initiate repairs of the malfunc-

tions because he had the technical insight to make decisions in technical matters because he

was perceived to be a representative of the technical management. A perception that was en-

hanced by the warranty engineers willingness to participate in fault finding and repairs. Fur-

of 45

thermore, the craft was continuously being repaired and there was confidence that the craft

would eventually become mechanically reliable.

Secondly, there was a deadline for bringing the craft into service that had an impact on the

decision to push the limits for when the craft was considered to be fully operational. That limit

was reached when the ship arrived in Bagenkop and the management acknowledged that it was

not possible to make the craft available for the charterer within the deadline.

These mechanical problems influenced the reaction of the crew in relation to the malfunctions

that occurred shortly after departure from Bagenkop. The alarms for cooling water low pres-

sure and high temperature on the lift fan engine were not considered to be an extraordinary

event because the master’s and mate’s mind-set and sensitivity towards alarms had been altered

during the events of the last four days. When the cooling water system malfunctioned, several

alarms on the cooling water system and lifting fan engine indicated that there was a technical

problem. The crew did not consider these alarms to be safety critical. Since the craft had left

the shipyard, the crew had been exposed to an abundance of true and false engine system

alarms and several technical breakdowns which made the crew consider the craft as being in a

permanently unstable condition. During arrival in Bagenkop, the focus of the crew was on

bringing the craft safely alongside and not on the failing cooling water system. It was never

considered an option that the malfunctioning cooling water system would or could cause a fire.

Considering the events in the days prior to the fire, the craft was not in a stable and seaworthy

condition. The national6 regulation assigns responsibility to the master regarding establishing

the seaworthiness of the craft, but that responsibility was not necessarily matched by the mas-

ter’s power to act. The master worked in an environment of distributed authority – between

the charterer, the ship management organisation, and the owners. In the continuous communi-

cation with the shore-based technical and commercial management, the master was subjected

to other forms of authority that challenged his perception of his own authority on board the

craft. Thereby, this accident highlights how the position of the master can be challenged in

relation to the management of the craft.

5.3 UMOE VENTUS’ robustness towards fire

UMOE VENTUS’s robustness towards the spread of a fire consisted of the structural fire pro-

tection (bulkheads, insulation, etc.), the functional fire protection (various types of equipment,

e.g. portable fire extinguishers) and the crewmembers’ operational capacity to fight the fire by

using the on-board procedures. These fire protection initiatives were based on the standards

set by the classification society and the Danish flag State regulation. Additional equipment to

the craft’s fire suppression capabilities was installed at the owner’s initiative, i.e. having several

different systems; a sprinkler system, a foam system and a gas system as fixed firefighting sys-

tems for the various compartments or having fire fighter equipment.

6 The merchant shipping act no. 75 of 17 January 2014, chapter 6, section 131.

of 45

All of the fire protection systems were based on the premise of the occurrence of certain fire

scenarios, e.g. engine room fires or fires inside the accommodation were potential sources of

fire that could be identified. It will be addressed in the following why these fire protection ini-

tiatives were inadequate to contain and control the fire scenario that occurred on 23 December

2015.

Structural and functional fire protection 5.3.1

Neither the manufacturer nor the authorities considered the lift fan compartments to be ex-

posed to a high risk of fire. Therefore, there was no regulatory or design requirement for the

lift fan compartments to be either constructed with insulated bulkheads for fire protection or

equipped with fire detectors and a fixed firefighting system. The ignition of the fire could hap-

pen unnoticed by the crew and could rapidly develop and spread from the port side lift fan

compartment because the open lift fan compartment provided large volumes of air to the fire,

and the surrounding bulkheads added combustible material to the fire. Subsequently, the fire

spread unhindered from the lift fan compartment upwards to the deck area and accommoda-

tion, and inwards to the diesel oil tank.

The fixed firefighting equipment (sprinkler, gas and foam installations) was designed to func-

tion within selected enclosed spaces (engine rooms and accommodation) and would therefore

not be effective in a fire scenario where the fire originated from the lift fan compartment and

spread to the open deck outside of the accommodation.

Once the fire had spread to the bulwark and the outside of the accommodation, the only op-

tion for the crewmembers was to utilize the fire hoses and portable extinguishers. After the

discovery of the fire, the crewmembers had little time to evaluate the situation because the fire

quickly became so intense that their immediate focus was on evacuating the craft because it

was apparent for them that it would be futile to start extinguishing the fire with the equipment

designed for a fire on deck. Furthermore, the smoke from the fire was perceived to be highly

toxic, which hindered free movement on the deck and influenced the decision to immediately

evacuate the craft without attempting to extinguish the fire.

The rapid development of the fire was made possible by the uninsulated carbon fibre rein-

forced plastic sandwich construction bulkheads that fuelled the fire and enabled it to spread

uncontrollably, which resulted in the structural collapse of the deck between the hulls. The use

of uninsulated carbon fibre reinforced plastic sandwich construction bulkheads will be elabo-

rated in the following.

Carbon fibre reinforced plastic sandwich bulkheads

UMOE VENTUS had been built with the basic design philosophy of a craft that could trans-

fer personnel and/or goods to offshore wind turbines in higher sea states than other similar

craft, while at the same time being an energy effective platform with very high speed and low

fuel consumption. The craft had been designed to have a distinct design profile as regards its

form, function, operative performance, choice of materials and components. To achieve these

of 45

structural and operational goals, the craft had, inter alia, been constructed with carbon fibre

reinforced plastic sandwich bulkheads.

As described in section 4.1.3, a general key aspect in the design of composite structures is the

ability to take advantage of the building material and utilize it to its operational limits. That

operational limit is dependent on the choice of material and manufacturing method. On

UMOE VENTUS the composites were sensitive to heat in relation to the structural integrity

of the deck and bulkhead and sensitive to a rapid development of fire because the composite

materials were combustible.

These characteristics of the composites led to requirements for equivalent solutions to meet

the protective characteristics that class A and B bulkheads provide according to the require-

ments in the national regulation7, which were based on constructing bulkheads of steel or other

equivalent material.

UMOE VENTUS had been constructed on the basis of DNV-GL classification standards for

small high-speed service craft for operation on offshore installations with additional require-

ments from the Danish Maritime Authority regarding, inter alia, the structural fire protection

inside the engine rooms. These standards required the engine room to be fitted with structural

fire protection only inside the compartment where it was imagined that the fire could originate.

One characteristic of having a fire-resistant bulkhead, such as steel, is that it offers a passive

structural protection with little sensitivity to variation in emergency scenarios, i.e. various types

of fire scenarios not previously imagined in the design process. Furthermore, it can structurally

contain fire for a long period, which provides time for the crew and passengers to evacuate the

craft without depending on the active use of extinguishing equipment. The purpose of a func-

tional fire protection system is to extinguish or contain fire based on conditional use. This means

that the functioning of the system is dependent on the mechanical reliability of the system and

that it is operationally applied in a correct manner at the correct time. With a lack of technical

and social redundancy,8 the functional fire protection systems are highly sensitive to changing

circumstances that are not within the designer’s imagination about possible emergency scenari-

os. Especially when the emergency situation evolves into a complex situation with multiple

simultaneous events, e.g. loss of power, breakdown of pumps, steam from the sprinklers scold-

ing the crew, etc. The sensitivity of such a system was described in the DMAIB accident report

about the fire on SEA GALE on 20 May 20149, which illustrated how a seemingly simple op-

erator error in connection with an emergency shutdown rendered the ship’s water mist system

ineffective.

7 Order no. 491 of 13 May 2014 on Notice B from the Danish Maritime Authority, the construction and equipment,

etc. of ships, chapter B II-2, Construction – Fire protection, fire detection and fire extinction, regulation 3 (Defini-tions).

8 Additional personnel with knowledge about how to operate the system.

9 SEA GALE – Fire on 20 May 2014 (www.dmaib.com).

of 45

The fire test of the bulkhead showed that the ignition temperature was relatively low, and the

sequence of events showed that the fire could rapidly spread. Building the craft in a combus-

tible carbon composite structure with a relatively low ignition temperature reduced the poten-

tial for controlling and limiting a fire to the crewmembers’ capacity to extinguish the fire by use

of the craft’s firefighting equipment and the strategies described in the safety management sys-

tem. The equipment and strategies were, however, based on the premise of fighting fires on a

ship constructed in a non-combustible material, which will be elaborated in the following.

The crew’s capacity to fight the fire 5.3.2

The crewmembers did not utilize the shipboard procedures in this specific emergency scenario

because the situation was not manageable with the procedural strategies described in UMOE

VENTUS’s safety management system. The DMAIB has previously addressed the fact that

static and prescriptive procedures can be found irrelevant by the crew in emergency situations

that are inherently dynamic10 and where there are problems with the practical aspects of how

to gain access to the procedure if it is stored in a computer which is not accessible, or how to

read the paper procedure on open deck with gale gusts in the midst of a fire. The emergency

procedures on UMOE VENTUS were not designed to be used in adverse environments and

under the cognitive pressure that the crew were exposed to. The stressful situation limited the

cognitive capabilities of the crew to such an extent that the procedure became useless.

The three crewmembers immediately felt overwhelmed by the intense fire and smoke and

therefore abandoned any attempt to fight the fire, including closing the quick closing valves,

activating any fire extinguishing system or using the fire hoses on deck. The main focus quickly

became to evacuate the craft.

This accident raises questions about the usefulness of procedures and operational practises that

are rooted in larger more robust ships where the timespan to assess, inform and act is longer.

Prolonging the decision to evacuate a small craft like UMOE VENTUS by taking all the pro-

cedural descriptions into account could result in a situation where an orderly evacuation of the

crew and passengers would be impossible.

CONCLUSIONS 6.

The fire on UMOE VENTUS on 23 December 2015 was caused by insufficient cooling of the

lift fan engine exhaust system, which ignited the exhaust muffler and/or the deck where it was

mounted in the open lift fan compartment on the port side. From the exhaust muffler and/or

deck the fire quickly spread outwards to the bulwark and accommodation and inwards to the

adjacent diesel oil tank. The insufficient cooling of the lift fan engine was likely caused by a

clogged sea chest strainer. There were several alarms on the cooling water system during arrival

and departure from Bagenkop, but the importance of the alarms was not acknowledged by the

10

Safety report – Proceduralizing marine safety, DMAIB (2016).

of 45

crew due to events that had unfolded during the preceding days which had created a tolerance

towards safety critical alarms.

Within 15 minutes the craft was engulfed in flames and drifted uncontrollably until it grounded

in the shallow waters north of the harbour of Bagenkop, Denmark, and was lost. After the

discovery of the fire, the crew had no other option than to evacuate the craft without any at-

tempt to fight the fire manually and/or by means of the craft’s fixed firefighting systems. The

overall aim of the investigation was thus to establish why the failing cooling water system led

to an uncontrollable fire that engulfed most of UMOE VENTUS within approximately 15

minutes after the fire was visually detected and resulted in a total loss of the craft.

UMOE VENTUS had been designed and constructed to be an energy effective platform with

high speed and low fuel consumption. The craft had been designed to have a distinct design

profile as regards its form, function, operative performance, choice of materials and compo-

nents. To achieve these structural and operational goals, the craft had, inter alia, been con-

structed with combustible carbon fibre reinforced plastic sandwich bulkheads. These compo-

site bulkheads had a relatively low ignition temperature which enabled the fire to rapidly spread

and engulf the craft within a short time span.

This accident illustrated that it can be problematic to change a ship’s construction from a non-

combustible material to a combustible material by designing equivalent solutions based on tra-

ditional functional fire protection strategies. It was found that the concept of building the craft

in a combustible carbon composite structure with a relatively low ignition temperature reduced

the craft’s robustness towards fire scenarios that were not considered during the design and

approval of the craft. This diminished the crew’s capacity to contain and control the fire by

utilizing the resources on board, which proved to be inadequate to fight the fire because they

had been designed to be used on a conventionally constructed ship. After the discovery of the

fire, the three crewmembers immediately felt overwhelmed by the intense fire and smoke and

therefore abandoned any attempt to fight the fire, and their main focus quickly became to

evacuate the craft. The strategies set out in the safety management system were not suitable for

handling the emergency situation that faced the crew because the strategies were based on a

type and size of ship with a robustness that would give the crew the capacity and time to con-

tain, control and assess the fire scenario. The fire on UMOE VENTUS shows the necessity of

rethinking the entire concept of the interaction between structural and functional fire protec-

tion, firefighting and evacuation when changing the underlying premise of having the ship con-

structed in a non-combustible material.

of 45

PREVENTIVE MEASURES TAKEN 7.

The DMAIB has received the following information from the shipyard about the preventive

measures taken in order to increase the robustness of existing and future craft:

“Umoe Mandal has carried out internal investigation in order to understand the chain of events and to, if

possible, establish the root cause of the accident. This investigation shows that the fire started in the port

Lift Fan room which is a part of the Surface Effect Ship system. This is the location of one of the two lift

fans that partly lifts the vessel out of the water. These rooms are classified as part of open deck spaces and

were not originally equipped with fire detectors and fixed firefighting equipment.

It is concluded that the fire was caused by lack of exhaust cooling from the port lift fan diesel engine provid-

ing power to the lift fan. The exhaust is normally cooled by means of seawater reducing temperatures from

approximately 450 degrees to 50 degrees. The exhaust passes through the bulkhead between the “lift fan

engine room” and the “lift fan room”. The passage through the bulk head is heat insulated. The exhaust

line ends in a silencer made in composite materials. This type approved silencer is designed to resist tempera-

tures up to 300 Degrees Celsius over a short period.

In order to increase robustness of the vessel, Umoe Mandal has implemented certain measures on the Sister

Ship Umoe Firmus currently operating in Wind Farms and on existing and future new builds:

A. Installed “Shut Down” switch on lift fan engines which will react on high exhaust temperatures (above

100°C). They come in addition to the switches originally installed which shut down engines based on

high fresh water cooling temperature.

B. Installed Multispectrum IR Flame detectors in lift fan rooms, connected to the Fire Alarm System.

C. The access to the lift fan area is simplified by installing a central locked hatch through deck.

D. Fit the vessels with two mobile “fog nails” used to penetrate composite panels with a nozzle connected

to the main fire seawater pumps – typically to be used if there is a fire in an area not covered by fixed

installations.

E. Installed a an additional Gas Fire Fighting system in the centre corridor between the lift fan rooms,

main engine rooms and generator rooms/water jet rooms. The objective is to cover areas not previously

covered.

F. Installed a CCTV Camera in the centre corridor between the two main engine rooms.

G. Updated the Ship Operating Manual to reflect the above measures. This includes system descriptions

and guidance to the use of these systems. When the DMAIB report is finalised the Ship Operating

manual will be reviewed again and updated to reflect all conclusions after the fire on Umoe Ventus. 2

major topics – typically relevant to Surface Effect Ships only - are already identified which will reduce

spread of fire and reduce risk to personnel when abandoning ship.

H. Installed on-line monitoring of alarms and running data from Ship Technical Control system directly

to the shipyard via satellite communications. This system will be used to support the crew when needed.

I. Carry out muster, firefighting and man-over-board exercises before delivery of vessels.”

Appendix 1

Appendix 2

Test Reg.no. 12

Danish Institute of Fire and Security Technology

Jernholmen 12, DK-2650 Hvidovre · Phone: +45 36 34 90 00 · E-mail: [email protected] · www.dbi-net.dk

D B I

Test report

Sandwich panels from ship

Name of client: Den Maritime Havarikommission

File no.: PFO10081A

Date: 2016-05-25

Pages: 11 Encl.: 9

Ref: MPA / DHL

08 Fall

Appendix 3

D B I

Page 2 of 11

File: PFO10081A Date: 2016-05-25

Client information

Client: Den Maritime Havarikommission

Address: Carl Jacobsens Vej 29

DK-2500 Valby

Denmark

The results relate only to the items tested. The test report should only be reproduced in extenso - in extracts

only with a written agreement with this institute.

Executive summery

The conclusion on ignition temperature is that the White panel can ignite in the smoke gas temperature range

of approx. 275-315 °C and the Untreated panel in the range of approx. 330-370 °C. This gas temperature should be measured 5mm above the surface using an ø1mm thermocouple. The investigation has been

conducted in the cone calorimeter.

“Ignition temperature” is a non-fundamental parameter that should be used with care. Ignition occurs when the right smoke gas temperature and the right smoke gas/oxygen ratio are present – this is never the same

in each test. Is either “correct” temperature or “correct” smoke gas/oxygen ratio missing - no ignition occurs.

Furthermore temperature is device dependent measurement.

DBI refers to the subsequent text which provide insight into the study forming the basis for this statement

D B I

Page 3 of 11

File: PFO10081A Date: 2016-05-25

Products Sandwich panels.

Description

Hull construction material for ships.

Manufacturer Not stated.

Purpose of tests In connection with fire accident of the ship named Umoe Ventus the client desired an investigation on the

ignition properties for the sampled products. The products have been subjected to the test procedures ISO

5660-1 and with additional ad hoc temperature measurements on the surface and above the surface of the product.

Sample 2016-05-11 DBI - Danish Institute of Fire and Security Technology received the following sample:

- 2 sandwich panels with white paint on one side with dimension 500 mm x 500 mm x 42.5 mm. One

panel was marked “AB-A7-KB40-A7 Skott LFR”. This panel type is designated “White panel” in this

report.

- 2 sandwich panels without paint with dimension 500 mm x 500 mm x 42.5 mm. One panel was marked “1-23-226-005”. This panel type is designated “Untreated panel” in this report.

All panels consisted of approx. 40 mm foam with approx. 1 mm glass fiber on both sides.

The weight per unit area of the White panel at 20°C (undried) was 8.4 kg/m2 at the state of receipt as determined by weight and measures of the sample.

The weight per unit area of the Untreated panel at 20°C (undried) was 7.9 kg/m2 at the state of receipt as

determined by weight and measures of the sample.

Conditioning 2015-05-13 the specimens were stored in a conditioning room with an atmosphere of relative humidity of 50

5% at a temperature of 23 2 C. The specimens were kept in this room until the tests were performed.

Test method The tests were performed in accordance with:

ISO 5660-1:2015 Reaction-to-fire tests – Heat release, smoke production and mass loss rate -

Part 1: Heat release rate (cone calorimeter method) and smoke production rate (dynamic measurement)

D B I

Page 4 of 11

File: PFO10081A Date: 2016-05-25

Test results 1.1 Test overview

Table 1 shows the overview of the performed tests with ignition results. The temperatures listed in Table 1 are average values in cases where more than one thermocouple was used. The ignition time and

temperatures are determined by the measuring point where a steep temperature rise is found which. The

temperatures were logged with a scan rate of 2 seconds. The found ignitions times by thermal measurement varies from the ISO 5660-1 shown in the following tables due the ISO 5660-1 data was logged with a scan

rate of 3 seconds.

Temperature measurements on and above the specimen were conducted with 1 mm sheathed K-type elements. Surface measurements were performed with the thermocouple place at an angle

of approx. 45 °C to the horizontal plane.

After test 9 it was decided to increase the number of temperature measurements and also measure the air

temperature above the specimen. Air temperatures measurements were performed with a distance of the thermocouple of approx. 5 mm to the surface.

Temperature graphs are shown in enclosure 1-5.

1.2 ISO 5660-1 results

Irradiance level: performed with varying irradiance levels (see tables) and all performed without spark igniter.

Orientation: Horizontal.

The retainer frame was used. In order to protect the foam of cut edges - which was assumed to have worse reaction fire properties than the glass fiber surface - a 13 mm ceramic fibre blanket was installed between the

specimen and the retainer frame outside the exposed area. The layer was compressed to approx. 4 mm.

Above the retainer frame a square perimeter of 13 mm ceramic fibre blanket was installed in order to reduce the heat flux of the steel sides of the frame retainer.

The specimen was laid onto one layer of 13 mm ceramic fibre blanket with a density of 65 kg/m³ taken from

DBI’s stock.

When testing the foam directly in test No. 18 and 20 - three layers of 13 mm ceramic fibre blanket were

used.

Volume flow was set to 24 ± 2 l/s.

All tests were performed with a distance of 60 mm between the cone and the specimens’ surface in order to

make room for thermal measurements.

The ISO 5660-1 test results where ignition occurred are shown in the following tables 2-5.

Graphs of heat release rate and smoke production rate for tests where ignition occurred are shown in

enclosure 6-9

D B I

Page 5 of 11

File: PFO10081A Date: 2016-05-25

Test no.

Surface

Heat flux (kW/m

2)

Amount of thermal couples on surface

Amount of thermal couples above surface

ISO 5660 measure ments

Ignition Ignition time (s)

Ignition tempe rature surface (°C)

Ignition tempera ture above specimen (°C)

Test 1 Untreated panel 10 1 0 no no - - -

Test 2 White panel 10 1 0 no no - - -


Test 4 White panel 20 1 0 no no - - -

Test 5 Untreated panel 30 1 0 no yes 248 439 -


Test 7 White panel 30 1 0 yes yes 158 454 -

Test 8 White panel 30 1 0 yes nej - - -

Test 9 White panel 30 1 0 yes yes 138 380 -

Test 10 White panel 35 2 2 yes yes 64 357 274



Test 13 Untreated panel 35 3 1 yes yes 78 384 367




Test 17 Untreated panel 30 2 2 yes no - - -

Test 18 Foam 30 2 2 yes no - - -

Test 19 Untreated panel 32.5 2 2 yes no - - -

Test 20 Foam 35 2 2 yes yes 758 521 378


Table 1

D B I

Page 6 of 11

File: PFO10081A Date: 2016-05-25

White panel

Test no. 7 9 Average

Flux level (kW/m2) 30 30 -

Exposed area As, (m2) 0.008836 0.008836 -

Calibration constant C 0.0476 0.0476 -

Mass before, mi (g) 84 85 85

Mass after, mf (g) 53 50 52

Mass loss, (g) 31 25 28

Flashing, tflash (s) - -

Ignition, tign(s) 159 138 148

All flaming ceased, text (s) 447 345 391

Test time, ttest (s) 1197 1746 -

Peak heat release rate, qmax

(kW/m²) 292 284 288

Total heat produced, THR (MJ/m²)

39 37 38

Average heat release, 3 min, q180 (kW/m²), after ignition

173 171 172

Average heat release, 5 min,

q300 (kW/m²), after ignition 116 109 113

Maximum Arhe, Marhe (kW/m²)

95 100 98

Peak smoke production rate,

RSP (m2/m2/s) 17 17 17

Total smoke produced, SA (m2/m²)

2537 2125 2331

Total smoke produced over the

non-flaming phase, SA,1 (m2/m²)

216 112 164


flaming phase, SA,2 (m2/m²) 2321 2013 2167

Table 2

D B I

Page 7 of 11

File: PFO10081A Date: 2016-05-25

White panel

Test no. 10 11 12 Average

Flux level (kW/m2) 35 35 35 -

Exposed area As, (m2) 0.008836 0.008836 0.008836 -

Calibration constant C 0.048 0.048 0.048 -

Mass before, mi (g) 86 86 86 86

Mass after, mf (g) 51 52 50 51

Mass loss, (g) 35 34 36 35

Flashing, tflash (s) - - - -

Ignition, tign(s) 63 78 75 72

All flaming ceased, text (s) 450 465 627 514

Test time, ttest (s) 1299 1041 1107 -


(kW/m²) 251 260 310 273


40 48 44 44


153 167 156 159


q300 (kW/m²), after ignition 106 130 107 114


122 120 128 123


RSP (m2/m2/s) 14 15 16 15


1731 2078 1675 1828


non-flaming phase, SA,1 (m2/m²)

15 20 42 26


flaming phase, SA,2 (m2/m²) 1716 2058 1633 1802

Table 3

D B I

Page 8 of 11

File: PFO10081A Date: 2016-05-25

Untreated panel

Test no. 13 14 15 21 Average

Flux level (kW/m2) 35 35 35 35 -

Exposed area As, (m2) 0.008836 0.008836 0.008836 0.008836 -

Calibration constant C 0.048 0.048 0.048 0.048 -

Mass before, mi (g) 83 83 84 83 86

Mass after, mf (g) 50 53 52 53 51

Mass loss, (g) 33 30 32 30 35

Flashing, tflash (s) - - - - -

Ignition, tign(s) 78 75 90 75 80

All flaming ceased, text (s) 636 261 633 1020 638

Test time, ttest (s) 897 846 1110 1023 -


(kW/m²) 281 309 291 258 285


51 30 43 54 45


178 152 157 172 165


q300 (kW/m²), after ignition 136 93 114 126 117


127 127 109 126 122


RSP (m2/m2/s) 19 21 19 18 19


2986 2032 2526 3267 2703

Total smoke produced over

the non-flaming phase, SA,1 (m2/m²)

22 20 52 22 29

Total smoke produced over

the flaming phase, SA,2 (m2/m²)

2965 2013 2474 3245 2674

Table 4

D B I

Page 9 of 11

File: PFO10081A Date: 2016-05-25

Foam

Test no. 20

Flux level (kW/m2) 35

Exposed area, As (m2) 0.008836

Calibration constant, C 0.048

Mass before, mi (g) 63

Mass after, mf (g) 11

Mass loss, (g) 51

Flashing, tflash (s) -

Ignition, tign(s) 758

All flaming ceased, text (s) 1208

Test time, ttest (s) 1316

Peak heat release rate, qmax (kW/m²) 233

Total heat produced, THR (MJ/m²) 55

Average heat release, 3 min, q180 (kW/m²), after ignition 168

Average heat release, 5 min, q300 (kW/m²), after ignition 136

Maximum Arhe, Marhe (kW/m²) 45

Peak smoke production rate, RSP (m2/m2/s) 12

Total smoke produced, SA (m2/m²) 4437

Total smoke produced over the non-flaming phase, SA,1 (m2/m²) 2841

Total smoke produced over the flaming phase, SA,2 (m2/m²) 1595

Table 5

For better understanding the nature of heat flux radiation Table 6 shows the average temperature of the

electrical cone element at the actual heat flux levels from the tests at a distance of 60 mm between cone and specimen.

Heat flux (kW/m2) Average temperature (°C) Distance to specimen (mm)

10 476 60

20 611 60

25 660 60

30 705 60

32.5 721 60

35 742 60

Table 6

D B I

Page 10 of 11

File: PFO10081A Date: 2016-05-25

Conclusion It was found that both the White panel and the Untreated panel could ignite at a flux level of 30 kW/m2 – however as Table 1 shows ignition did not always occur meaning 30 kW/m2 must the lower limit of ignition for

both panel types. As shown in the temperature curves in enclosure 1-6 unignited specimen follow more or less the same temperature curves as the ones that did ignite.

At 35 kW/m2 all panels ignited during the tests. The White panel ignited in average after 72 seconds. For the White panel the average surface temperature at ignition was found to be 374 °C and the average

temperature above the specimen to be 294 °C.

At the same flux level the Untreated panel ignited in average after 80 seconds. For the Untreated panel the

average surface temperature at ignition was found to be 395 °C and the average temperature above the specimen to be 343 °C.

Test 15 shows that a prolonged ignition time apparently results in a higher surface temperature whereas the

air temperature is on the level with the average value. In regard to surface temperature this is also the case with test 7 and 9.

It also seems that the surface temperature at ignition found at the 30 kW/m2 are higher than for the 35 kW/m2 tests – though ignition data for 30 kW/m2 is scarce.

In regard to ignition the White panel was found to be only slightly worse than the Untreated panel.

Tests without pilot burner or spark ignition tend to vary much more in ignition time than tests with pilot burner or spark ignition. Ignition occurs when the right smoke gas temperature and the right smoke

gas/oxygen ratio are present – this is never the same in each test. The tests where ignition did not occur were due to the fact that the temperature and ratio were never there. It is also possible that the variation of

fire retardant in the panels causes the different ignition behaviour - for instance one test at 32.5 kW/m2

showed no ignition – based on former results it was expected to ignite.

Two tests were conducted with the foam directly exposed. The test at 30 kW/m2 showed no ignition as most as the other tests show. At 35 kW/m2 the ignition came significantly later compared to the other tests. It was

also noted that the smoke colour was lighter than observed in the other tests indicating a higher amount of less volatile gas compounds or water. The surface temperature at ignition was found to be 521 °C and the

temperature above the specimen to be 378 °C.

As the occurrence of ignition is correlated to the smoke gas temperature and the smoke gas/oxygen ratio the

surface temperature is not a good indication for determining ignition. The smoke gas temperature is a better parameter for determining ignition.

The conclusion on ignition temperature is that the White panel can ignite in the smoke gas temperature range of approx. 275-315 °C and the Untreated panel in the range of approx. 330-370 °C.

In regard to heat release rate the results no significant difference was found between the White and the

Untreated panel.

D B I

Page 11 of 11

File: PFO10081A Date: 2016-05-25

Statement These test results relate only to the behaviour of the product under the particular conditions of the test, and

they are not intended to be the sole criterion for assessing the potential fire hazard of the product in use.

Dan Lauridsen

M.Sc. (Civ.Eng.)

Martin Pauner

M.Sc.Civ.Eng

Den Maritime Havarikommission Carl Jacobsens Vej 29

DK-2500 Valby

Denmark

D B I

Enclosure 1 of 9

File: PFO10081A Date: 2016-05-25

D B I

Enclosure 2 of 9

File: PFO10081A Date: 2016-05-25

D B I

Enclosure 3 of 9

File: PFO10081A Date: 2016-05-25

D B I

Enclosure 4 of 9

File: PFO10081A Date: 2016-05-25

D B I

Enclosure 5 of 9

File: PFO10081A Date: 2016-05-25

D B I

Enclosure 6 of 9

File: PFO10081A Date: 2016-05-25

D B I

Enclosure 7 of 9

File: PFO10081A Date: 2016-05-25

D B I

Enclosure 8 of 9

File: PFO10081A Date: 2016-05-25

D B I

Enclosure 9 of 9

File: PFO10081A Date: 2016-05-25

marine accident report september 2016 - dmaib.dk ventus - fire on... · marine accident report...

Documents